Environ. Sci. Technol. 2002, 36, 996-1003
A Mass Balance Model Describing Multiyear Fate of Organochlorine Compounds in a High Arctic Lake PAUL A. HELM,† M I R I A M L . D I A M O N D , * ,†,‡ R A Y S E M K I N , § WILLIAM M. J. STRACHAN,§ CAMILLA TEIXEIRA,§ AND DENNIS GREGOR# Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto, Ontario, M5S 3E5, Canada, Department of Geography, University of Toronto, 100 St. George Street, Toronto, Ontario, M5S 3G3, Canada, National Water Research Institute, Canada Centre for Inland Waters, Burlington, Ontario, L7R 4A6, Canada, and MDA Consulting Limited, Unit 6A, 35 Crawford Cres., Campbellville, Ontario, L0P 1B0, Canada
Data collected over a 3-year study of a high arctic watershed and lake are used to understand the fate of organochlorine compounds (OCs) and form the basis of a mass balance contaminant fate model. The model uses the fugacity/aquivalence approach to describe OC dynamics between air, stream inflows and outflow, the water column, and surficial sediments. The steady-state model results indicate that stream inflows contributed from 96 to >99% of total chemical loadings, but 57-98% of total loadings were lost from the lake via the outlet, the percentage of which is controlled by the hydrologic regime of the high arctic lake. Conversely, only 0.4-3.4% of loadings were retained within the sediments due to the high export rate, minimal scavenging from the water column and low organic carbon fraction of the sediments. Using the unsteady-state model, which includes year-round processes, degradation was estimated to account for losses of 7-32% for the more persistent OCs and 42-50% for the less persistent OCs (R-HCH, γ-HCH, and endosulfan I). If loadings were eliminated, water column concentrations would decline with half-lives 2 m that is maintained until mid-June when the next annual cycle begins (12, 27). The area of open water available for air-water exchange and atmospheric deposition processes and the volume of unfrozen lake water were adjusted accordingly throughout the year, with approximately 10% of the lake volume occupied by ice at maximum thickness. Concentration functions for OCs in streams in the unsteady-state model were determined using nonlinear and linear regression techniques using measured concentrations and day as variables. These equations accounted for the preferential elution of more soluble OCs such as R- and VOL. 36, NO. 5, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
997
TABLE 2. Mean OC Concentrations and Ranges in Inflow Streams, Mean Concentrations ( Standard Deviations in Amituk Lake, the Outflow, and Alert Air, and Amituk Lake Sediment Concentrations inflow streams (pg L-1) compound meana R-HCH β-HCH γ-HCH dieldrin Endo I HepEx ∑DDT ∑PCBc
1470 39 376 71 137 61 44 576
1993 range
meana
n.d.-5429 9-109 5-1818 16-171 n.d.-477 6-129 n.d.-125 n.d.-1438
602 29 212 95 87 47 19 899
lake (pg L-1)b
1994 range
1993
1992 20 m
3m
20 m
1994 20 m
3m
40 m
6-2969 1963 ( 1018 1164 ( 322 1090 ( 261 677 ( 134 564 ( 86 646 ( 122 n.d.-101 48 ( 18 30 ( 10 41 ( 39 32 ( 14 30 ( 12 31 ( 10 n.d.-863 351 ( 155 191 ( 58 193 ( 120 191 ( 63 153 ( 29 163 ( 45 5-255 60 ( 44 51 ( 8 56 ( 22 86 ( 16 80 ( 27 81 ( 22 n.d.-408 45 ( 66 45 ( 23 3.9 ( 5.4 43 ( 36 20 ( 11 20 ( 9 7-130 44 ( 25 64 ( 7 63 ( 27 41 ( 13 38 ( 19 47 ( 12 n.d.-58 182 ( 145 39 ( 13 40 ( 37 8.6 ( 4.5 36 ( 16 35 ( 12 117-1706 1380 ( 663 417 ( 155 305 ( 115 421 ( 244 400 ( 190 294 ( 108
outflow (pg L-1)d
air (pg m-3)e
compound
1993
1994
1993
1994
R-HCH β-HCH γ-HCH dieldrin Endo I HepEx ∑DDT ∑PCBc
1383 ( 424 63 ( 33 194 ( 111 33 ( 33 69 ( 46 63 ( 15 40 ( 28 462 ( 266
854 ( 313 27 ( 11 255 ( 111 111 ( 35 71 ( 62 51 ( 19 9(4 487 ( 142
49.9 ( 26.4 0.4 ( 0.2 6.3 ( 3.0 1.7 ( 1.0 4.4 ( 3.2 1.6 ( 0.8 0.4 ( 0.2 39.5 ( 29.7
41.8 ( 15.7 0.1 ( 0.1 7.0 ( 4.2 1.4 ( 0.6 4.4 ( 2.8 1.5 ( 0.6 0.8 ( 0.4 32.1 ( 12.0
sediment (ng g -1)f 1992 0.35 0.01 0.55 1.18 0.10 1.67 12.2
a Weighted by stream contributions to total discharge. b n ) 12 in 1992, n ) 9 in 1993, n ) 11 in 1994. c Excludes outlier on Jun 16/93. 7 in 1993, n ) 17 in 1994. e Summarized in refs 6 and 7. f References 24 and 36. n.d. ) not detected.
d
n)
TABLE 3. Selected Physical-Chemical Properties of OCs Included in Model Calculations compound
MW (g mol-1)
solubilitya CL (mg L-1)
vapor pressureb PL (Pa)
Henry’s law constantc H (Pa m3 mol-1)
log KOWa
degradationd t1/2 (y)
R-HCH β-HCH γ-HCH dieldrin Endosulfan I HepEx ∑DDTe ∑PCBe
290.85 290.85 290.85 380.93 406.95 389.2 330.2 302.3
33.3 66.0 53.5 6.5 0.68 7.4 0.49 0.479 f
0.02188 0.0264g 0.005864 0.0006127 0.0003923 0.054g 0.0000779 0.00985h
0.1235i 0.116 0.07554i 0.036 0.23 2.8 0.02 1.261
3.81 3.8 3.7 5.2 3.62 5.0 6.05 6.1f
0.90j 12.6k 1.26j 22.0 0.05 5.71j 20.0 12.6l
a From ref 30 unless noted. b Calculated from ref 31 @ 275 K, unless noted. c Calculated from P /C unless noted. d From refs 30 and 32 unless L L noted. Selected values were doubled to adjust for the “Q-10 rule”. e Properties for ∑DDT and ∑PCB are averages weighted according to the occurrence of DDT, DDE, and DDE for ∑DDT and PCB homologues for ∑PCB in Amituk Lake and streams. f Weighted average of homologue values from ref 32. g From ref 30. h Weighted average of homologue values from ref 33 calculated @ 275 K. i Calculated from ref 34 @ 275 K. j Determined from model calibration. k From ref 30, class 9, doubled. l From ref 32, class 9, doubled.
γ-HCH from the snowpack and later elution of less soluble compounds such as ∑PCB and ∑DDT (11, 28). Daily chemical loadings were then calculated as the product of stream discharge and estimated daily chemical concentrations for each stream and then summed for all streams. Approximately 40-50% of annual precipitation occurs as rainfall during summer (18). Chemical loadings due to rain dissolution and wet deposition for the unsteady-state model were estimated by normally distributing rainfall beginning in mid-June, peaking in early August, and ending in late September (16) and incorporating this into model D-values. Air concentrations (Table 2) measured at Alert, NT, from June to August 1993 and 1994 were used in the models (summarized in Halsall et al. (7) and Stern et al. (6)). Although Cornwallis Island is ∼1060 km south of Alert, the arctic air mass is considered homogeneous over large regions of the Canadian Arctic (29). Table 3 lists selected physical-chemical properties of modeled OCs, corrected to 2 or 4 °C where possible for Henry’s Law constants (34) or vapor pressures (31, 33). Properties of ∑DDT and ∑PCB were obtained by selecting properties of individual compounds (DDT, DDE, DDD; average PCB homologue values) and weighting them ac998
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 5, 2002
cording to their abundance in Amituk Lake and streams. As a first approximation, the degradation or transformation halflives of the more persistent chemicals (e.g., PCBs, DDT, and dieldrin) were taken from Mackay et al. (30, 32), which pertain to temperate regions with average water temperatures of 10-15 °C. These values were doubled to account for the temperature-related decrease in biological degradation rates under arctic conditions (approximately 0-5 °C) according to the “Q-10 rule”. Rates of photolysis were considered negligible due to ice cover, and hydrolysis rates for these chemicals in water are low (e.g. R- and γ-HCH (35)), with the exception of endosulfan I. Degradation half-lives for the more reactive OCs were estimated through model calibration as discussed below. Sediment concentrations used to initialize the unsteadystate model were those reported in the surface slice (1 cm) of a sediment core taken from Amituk Lake in 1992 (24, 36). Endosulfan I was not analyzed for, but a value of 0.01 ng/g was assumed. This represents the limit of detection of most components in ref 24. The unsteady-state model was calibrated by comparison of estimated and measured lake concentrations of R-HCH for the summer of 1994 since this was the most abundant
FIGURE 1. Comparison of modeled (steady-state) and measured mean organochlorine compound concentrations in Amituk Lake water in 1993 and 1994. Measured concentrations were determined for the summer months at 20 m depth. Error bars represent ( one standard deviation about the measured mean. chemical and the most detailed measurements of stream chemistry, hydrology, and chemical concentrations were made in 1994. The enantiomer ratio (ER) of R-HCH was used to trace the inflow and mixing of freshly deposited, more racemic R-HCH from snowpack runoff with the more degraded R-HCH in lake water (37). Calibration was limited to parametrization of the time-dependent throughflow fraction. Based on R-HCH concentrations and ERs at 3 m depth and in the outflow at peak melt, it was determined that 85-100% of water at this depth consisted of inflow water (37) and thus the throughflow fraction was set at 0.8-0.9 from June 22-28. The highest total inflow rates occurred from June 29-July 4 during which time the R-HCH concentration and ER increased at 20 m depth, suggesting that a high proportion of the meltwater mixed with the water column and the throughflow fraction was set to 0.1. After this event discharges declined, and since partial ice-cover remained, the throughflow fraction was raised and allowed to gradually decline until ice-off. Under this mixing regime, the model reproduced the increases in concentrations observed for R- and γ-HCH at 20 m during peak melt in 1994.
Results and Discussion Steady-State Model. The steady-state model is a useful tool for identifying and investigating important contaminant fate processes and parameters which pertain to the short arctic summer. Although several parameters varied temporally, average constant values were used. The performance of the model was evaluated by comparing predicted lake concentrations to measured mean concentrations ( one standard deviation (Table 2) for 1993 and 1994 (Figure 1). Model estimates of bulk water column concentrations were generally within one standard deviation of the measured mean for each of the eight OC compounds or compound classes. Agreement between predicted and measured mean outflow concentrations (not shown) was similar to lake values. Mass outflow flux of chemical agreed to within ( 20-30% of measured mass outflow for 1993 and 1994. These results indicate that the steady-state model reasonably describes the overall fate of these OCs in Amituk Lake. Sediment concentrations predicted by the model could not be compared to published values for Amituk Lake (24, 36) as the measured chemical inputs were from the 1992-1994 melt seasons, whereas the median age of the surface sediment slice (1 cm) was 1980 (24). Contaminant fluxes into, within, and out of the lake were compared to identify the important loading and loss pro-
FIGURE 2. Comparison of relative losses, by process, of organochlorine compounds from Amituk Lake as estimated by the steadystate model for the summer of 1994. cesses in Amituk Lake. The inflow of OCs with meltwater from the snowpack was the dominant loading process, accounting for 96 to >99% of total annual loadings. Direct atmospheric loadings to the lake were minor, with 0.1-2.6% by absorption,
